Category: Science & Tech

That’s what I wrote about in a new story today. Chiplets—the new chipmaking approach that breaks down chips into independent modules to reduce design costs and improve computing performance—can help China develop more powerful chips despite US government sanctions that prevent Chinese companies from importing certain key technologies.

Outside China, chiplets are one of the alternative routes that the semiconductor industry could take to improve chip performance cost-effectively. Instead of endlessly trying to cram more transistors into one chip, the chiplet approach proposes that the functions of a chip can be separated into several smaller devices, and each component could be easier to make than a powerful single-piece chip. Companies like Apple and Intel have already made commercial products this way. 

But within China, the technology takes on a different level of significance. US sanctions mean that Chinese companies can’t purchase the most advanced chips or the equipment to make them, so they have to figure out how to maximize the technologies they have. And chiplets come in handy here: if the companies can make each chiplet to the most advanced level they are capable of and assemble these chiplets into a system, it can act as a substitute for more powerful cutting-edge chips.

The technology needed to make chiplet is not that new. Huawei, the Chinese tech giant that has a chip-design subsidiary called HiSilicon, experimented with its first chiplet design product in 2014. But the technology became more important to the company after it was subject to strict sanctions from the US in 2019 and couldn’t work with foreign factories anymore. In 2022, Huawei’s then chairman, Guo Ping, said the company was hoping to connect and stack up less advanced chip modules to keep the products competitive in the market. 

Currently, there’s a lot of money going into the chiplet space. The Chinese government and investors have recognized the importance of chiplets, and they are pouring funding into academic projects and startups.

Particularly, there’s one Chinese city that has gone all-in on chiplets, and you very likely have never heard its name: Wuxi (pronounced woo-she). 

Halfway between Shanghai and Nanjing, Wuxi is a medium-sized city with a strong manufacturing industry. And it has a long history in the semiconductor sector: the Chinese government built a state-owned wafer factory there in the ’60s. And when the government decided to invest in the semiconductor industry by 1989, 75% of the state budget went into the factory in Wuxi.

Full Transcript 

Laurel Ruma: From MIT Technology Review, I’m Laurel Ruma, and this is Business Lab, the show that helps business leaders make sense of new technologies coming out of the lab and into the marketplace.

Our topic today is blockchain. Technology has changed how money moves around the world, but the opportunity and value from distributed ledger technology is still in its early days. However, deploying on a large scale openly and securely should move it along quickly.

Two words for you: building innovation.

My guest is Suresh Shetty, who is the chief technology officer at Onyx by J.P.Morgan at JPMorgan Chase.

This podcast is produced in association with JPMorgan Chase.

Welcome, Suresh.

Suresh Shetty: Thank you so much, Laurel. Looking forward to the conversation.

Laurel: So to set the context of this conversation, JPMorgan Chase began investing in blockchain in 2015, which as we all know, in technology years is forever ago. Could you describe the current capabilities of blockchain and how it’s evolved over time at JPMorgan Chase?

Suresh: Absolutely. So when we began this journey, as you mentioned, in 2015, 2016, as any strategy and exploration of new technologies, we had to choose a path. And one of the interesting things is that when you’re looking at strategic views of five, 10 years into the future, inevitably, there needs to be some course correction. So what we did in JPMorgan Chase was we looked at a number of different lines of inquiry, and in each of these lines of inquiries, our focus was trying to be as inclusive as possible. So what we mean by that is that we actually weighted ubiquity in terms of who can use the technology, who was trying to use the technology over technology superiority. Because eventually, our feeling was that the network effect, the community effect of ubiquity, actually overcomes any technology challenges that a person or a firm might have.

Now, I think that a very relevant example is the Betamax-VHS example. It’s a bit dated but I think it really is important in this type of use case. So as many of you know, Betamax was a superior technology at the time and VHS was much more ubiquitous in the marketplace. And over time, what happened was that people gravitated, firms gravitated towards that ubiquity over the superiority of the technology that was in Betamax. And similarly, that was our feeling too in terms of blockchain in general and specifically the path that we took, which was in and around the Ethereum ecosystem. We felt that the Ethereum ecosystem had the largest developer community, and we thought over time, that was where we needed to focus in on.

So I think that that was our journey to date in terms of looking, and we continue to make those decisions in terms of collaboration, inclusiveness, as opposed to just purely looking at technology itself.

Laurel:And let’s really focus on those efforts. In 2020, the firm debuted Onyx by J.P.Morgan, which is a blockchain-based platform for wholesale payment transactions. Could you explain what wholesale payment transactions are and why they’re the basis of Onyx’s mission?

Suresh: Absolutely. Now, it was interesting. My background is that I came from the markets world and markets is really involved in front office trading, investment banking and so forth, and eventually, went over to the payments world. And if you juxtapose the two, it’s actually very interesting because initially, people feel that the market space is much more complicated, much more exciting than payments, and they feel that payments is a relatively straightforward exercise. You’re moving money from point A to point B.

What actually happens is actually, payments is much more complicated, especially from a transactional perspective. So what I mean by that is that if you look at markets, what happens is if you do a transaction, it flows through. If there’s an error, what you do is that you correct the initial transaction, cancel it, and put in a new transaction. So all you do is that there’s a series of cancel corrects, all of which are linked together by the previous transaction, so there’s a daisy chain of transactions which are relatively straightforward and easy to migrate upon.

But if you look at the payments world, what happens is that you have a transaction, it flows through. If there’s an error, you hold the transaction, you correct it, and then keep going. Now, if you think about it from a technology perspective, this is a lot more complicated because what you have to do is you have to keep in mind the state engine of the transactional flow, and you have to store it somewhere, and then you have to constantly make sure that as it flows to the next unit of work, it actually is not only referenced but it actually has the data and transactionality from the previous unit of work. So a lot more complicated.

Now, from a business perspective, what cross-border payments or wholesale payments involved is that, as I mentioned, you’re moving money from point A to point B. In an ideal fashion, and I’ll give you an example. Since I’m in India, in an ideal example, we would move money from JPMorgan Chase to State Bank of India, and the transaction is complete, and everybody is happy. And in between that transaction, we do things like a credit check to make sure that the money that is being sent, there’s money in the account of the sender. We need to make sure that the receiver of the account has a valid bank account, so you need to do that validation, so there’s a credit check. Then on top of that, you do a sanctions check. A sanctions check means that we are evaluating whether the money is being moved to a bad actor, and if it is, we stop the transaction and we inform the relevant parties. So it looks relatively straightforward in an idealized version.

But what if an AI could learn like a baby? AI models are trained on vast data sets consisting of billions of data points. Researchers at New York University wanted to see what such models could do when they were trained on a much smaller data set: the sights and sounds experienced by a single child learning to talk. To their surprise, their AI learned a lot—thanks to a curious baby called Sam. 

The researchers strapped a camera on Sam’s head, and he wore it off and on for one and a half years, from the time he was six months old until a little after his second birthday. The material he collected allowed the researchers to teach a neural network to match words to the objects they represent, reports Cassandra Willyard in this story. (Worth clicking just for the incredibly cute pictures!) 

WAI KEEN VONG

This research is just one example of how babies could take us a step closer to teaching computers to learn like humans—and ultimately build AI systems that are as intelligent as we are. Babies have inspired researchers for years. They are keen observers and excellent learners. Babies also learn through trial and error, and humans keep getting smarter as we learn more about the world. Developmental psychologists say that babies have an intuitive sense of what will happen next. For example, they know that a ball exists even though it is hidden from view, that the ball is solid and won’t suddenly change form, and that it rolls away in a continuous path and can’t suddenly teleport elsewhere. 

Researchers at Google DeepMind tried to teach an AI system to have that same sense of “intuitive physics” by training a model that learns how things move by focusing on objects in videos instead of individual pixels. They trained the model on hundreds of thousands of videos to learn how an object behaves. If babies are surprised by something like a ball suddenly flying out of the window, the theory goes, it is because the object is moving in a way that violates the baby’s understanding of physics. The researchers at Google DeepMind managed to get their AI system, too, to show “surprise” when an object moved differently from the way it had learned that objects move.

Yann LeCun, a Turing Prize winner and Meta’s chief AI scientist, has argued that teaching AI systems to observe like children might be the way forward to more intelligent systems. He says humans have a simulation of the world, or a “world model,” in our brains, allowing us to know intuitively that the world is three-dimensional and that objects don’t actually disappear when they go out of view. It lets us predict where a bouncing ball or a speeding bike will be in a few seconds’ time. He’s busy building entirely new architectures for AI that take inspiration from how humans learn. We covered his big bet for the future of AI here.

The AI systems of today excel at narrow tasks, such as playing chess or generating text that sounds like something written by a human. But compared with the human brain—the most powerful machine we know of—these systems are brittle. They lack the sort of common sense that would allow them to operate seamlessly in a messy world, do more sophisticated reasoning, and be more helpful to humans. Studying how babies learn could help us unlock those abilities. 

Deeper Learning

This robot can tidy a room without any help

Robots are good at certain tasks. They’re great at picking up and moving objects, for example, and they’re even getting better at cooking. But while robots may easily complete tasks like these in a laboratory, getting them to work in an unfamiliar environment where there’s little data available is a real challenge.

Immediate reactions to the government decision have been predictable. Some environmental organizations hailed the announcement as a much-needed course correction, arguing that it could help the US meet its global climate commitments. Industry trade groups, in turn, have attacked the decision. They insist it’s a counterproductive way to cut greenhouse-gas emissions, and one that will undermine the nation’s energy security at a moment of growing geopolitical volatility. 

Who is right? Turns out we are asking the wrong question.

What is important is not the absolute emissions associated with any given cargo ship full of LNG that departs from the US, the largest exporter of the product. Rather, when the fuel is exported, the net climate impact depends on what it replaces in the importing country, and whether realistic alternatives produce more or less greenhouse gas.

Consider this: The Russian war on Ukraine spurred dramatic growth in US exports of LNG to Europe. That gas was used primarily in the power sector to keep lights and heat on. In a parallel universe that did not see Russian aggression, the likely scenario would be a Europe that continued to purchase gas from Russia. Yet, as evidence shows, Russian natural gas is associated with higher methane emissions compared with the US natural-gas supply chain. That’s mainly because of Russia’s particularly leaky natural-gas infrastructure, which allows vast amounts of the potent greenhouse gas to escape into the atmosphere. In this context, replacing piped Russian gas with US LNG likely reduced overall carbon emissions, even with the added emissions from shipping the fuel across an ocean. 

Or let’s take another example: US LNG exports to India are first used for applications such as fertilizer manufacturing or heavy industry, and only then in the power sector. This is because solar energy is the cheapest form of power generation in India. In addition, coal plants produce the bulk of electricity generation, thanks in part to subsidies for the sector.

Given all this, there’s no scenario in India where high-priced LNG imports can compete with coal or crowd out lower-carbon renewables. So here, too, the fuel almost certainly won’t increase overall emissions from the power sector.

None of this is to say that US LNG always reduces emissions around the world. Indeed, the entire point of the above examples is that the climate impact of the fuel depends on a variety of factors and must be evaluated on a country-by-country basis. In addition, whether or not US LNG reduces emissions on net may change over time as countries decarbonize.

There is a legitimate debate to be had about the long-term impact of US LNG exports, and whether—or under what scenarios—these exports are compatible with global climate agreements.

Over the past decade, the main way that natural gas has helped reduce emissions is by replacing dirtier coal-burning power plants. But how much longer the fuel can continue to help depends on our emissions and warming trajectories.

Human babies are far better at learning than even the very best large language models. To be able to write in passable English, ChatGPT had to be trained on massive data sets that contain millions upon millions of words. Children, on the other hand, have access to only a tiny fraction of that data, yet by age three they’re communicating in quite sophisticated ways.

A team of researchers at New York University wondered if AI could learn like a baby. What could an AI model do when given a far smaller data set—the sights and sounds experienced by a single child learning to talk?

A lot, it turns out. This work, published in Science, not only provides insights into how babies learn but could also lead to better AI models. Read the full story.

—Cassandra Willyard

The next generation of mRNA vaccines is on its way

Japan recently approved a new mRNA vaccine for covid, and this one is pretty exciting. Just like the mRNA vaccines you know and love, it delivers the instructions for making the virus’s spike protein. But here’s what makes it novel: it also tells the body how to make more mRNA. Essentially, it’s self-amplifying.

These kinds of vaccines offer a couple of important advantages over conventional mRNA vaccines, at least in theory. The dose can be much lower, and it’s possible that they will induce a more durable immune response.